Mapping infectious disease landscapes: unmanned aerial … · 2015-01-16 · Mapping infectious disease landscapes: unmanned aerial vehicles and epidemiology Kimberly 1 M. Fornace1,

TREPAR-1317; No. of Pages 6

Mapping infectious diseaselandscapes: unmanned aerial vehiclesand epidemiologyKimberly M. Fornace1, Chris J. Drakeley1, Timothy William2,3,4, Fe Espino5, andJonathan Cox1

1 Faculty of Infectious and Tropical Diseases, London School of Hygiene and Tropical Medicine, London, UK2 Infectious Diseases Society Sabah, Menzies School of Health Research Clinical Research Unit, Kota Kinabalu, Sabah, Malaysia3 Infectious Diseases Unit, Clinical Research Centre, Queen Elizabeth Hospital, Kota Kinabalu, Sabah, Malaysia4 Sabah Department of Health, Kota Kinabalu, Sabah, Malaysia5 Research Institute for Tropical Medicine, Department of Health, Filinvest, Alabang, Muntinlupa City, Philippines

Opinion

The potential applications of unmanned aerial vehicles(UAVs), or drones, have generated intense interestacross many fields. UAVs offer the potential to collectdetailed spatial information in real time at relatively lowcost and are being used increasingly in conservation andecological research. Within infectious disease epidemi-ology and public health research, UAVs can providespatially and temporally accurate data critical to under-standing the linkages between disease transmission andenvironmental factors. Using UAVs avoids many of thelimitations associated with satellite data (e.g., long re-peat times, cloud contamination, low spatial resolution).However, the practicalities of using UAVs for field re-search limit their use to specific applications and set-tings. UAVs fill a niche but do not replace existingremote-sensing methods.

Applications of UAVsIncreasing attention has been focused on the potential usesof UAVs. UAVs have been used for various civilian purposesranging from law enforcement, fire fighting, and parceldelivery to wildlife population monitoring [Handwerk, B.(2013) Five surprising drone uses (besides Amazon deliv-ery). National Geographic (http://news.nationalgeographic.-com/news/2013/12/131202-drone-uav-uas-amazon-octocop-ter-bezos-science-aircraft-unmanned-robot/)] [1,2]. UAVsoffer the potential to collect detailed geospatial informationin real time at relatively low cost. UAVs can also be aneffective method of monitoring situations too dangerous orcostly for traditional aerial surveys, such as mapping forestfires and ice floes in the Arctic or conducting antipoachingpatrols [Dillow, C. (2014) Drones are invading the Arctic!CNBC (http://www.cnbc.com/id/101417956)] [3]. Theseadvantages have led to the application of UAVs for ecologicalresearch studies evaluating land use and cover change and

1471-4922/

� 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.pt.2014.09.001

Corresponding author: Fornace, K.M. ([email protected]).Keywords: geographic information system; unmanned aerial vehicle; drone; spatialepidemiology; malaria; Plasmodium knowlesi.

conducting aerial surveys of large wild animals such asdugongs, rhinoceros, and orangutans [3–7]. Additionally,UAVs have been used in agriculture to monitor vegetationlevels, crop growth, and distribution of weeds [8,9].

There are also numerous potential applications for UAVsin public health. UAVs can be used to locate people andmonitor human population movements of nomadic andmigrant groups to allow targeting of surveillance and publichealth interventions [10]. UAVs have also been used tofacilitate access to and sample collection from remote loca-tions. For example, a UAV was developed to allow thetransportation of test samples from remote rural clinics tonational laboratories in South Africa [11]. UAVs can also beused for disaster management and emergency relief opera-tions to monitor situations as well as to deliver medicalsupplies to inaccessible or dangerous locations. During theaftermath of Typhoon Haiyan in the Philippines, UAVs wereused by aid organisations to assess the extent of the typhoondamage and plan relief measures and reconstruction [Klap-tocz, A. (2014) Mapping the Philippines after TyphoonHaiyan. Drone Adventures (http://www.droneadventure-s.org/2014/05/07/mapping-the-philippines-after-typhoon-haiyan/)]. Aid organisations have also started piloting theuse of UAVs to deliver medical supplies to areas inaccessibleby road in Haiti, the Dominican Republic, and Lesotho[Hickey, S. (2014) Humanitarian drones to deliver medicalsupplies to roadless areas. The Guardian (http://http://www.theguardian.com/world/2014/mar/30/humanitarian-drones-medical-supplies-no-roads-technology)].

UAVs can also be used to collect other types of environ-mental data of public health relevance. Environmental fac-tors such as radiation and air pollution vary spatially, withimportant consequences for human health. Monitoringequipment has been fitted to UAVs to measure levels ofenvironmental toxins and pollutants [12,13]. Further appli-cations could include mapping health infrastructure, suchas water and sanitation systems and locations of healthfacilities.

Within infectious disease epidemiology, UAVs provide anew alternative to collect detailed georeferenced informa-tion on environmental and other spatial variables

Trends in Parasitology xx (2014) 1–6 1

http://news.nationalgeographic.com/news/2013/12/131202-drone-uav-uas-amazon-octocopter-bezos-science-aircraft-unmanned-robot/http://news.nationalgeographic.com/news/2013/12/131202-drone-uav-uas-amazon-octocopter-bezos-science-aircraft-unmanned-robot/http://news.nationalgeographic.com/news/2013/12/131202-drone-uav-uas-amazon-octocopter-bezos-science-aircraft-unmanned-robot/http://www.cnbc.com/id/101417956http://www.droneadventures.org/2014/05/07/mapping-the-philippines-after-typhoon-haiyan/http://www.droneadventures.org/2014/05/07/mapping-the-philippines-after-typhoon-haiyan/http://www.droneadventures.org/2014/05/07/mapping-the-philippines-after-typhoon-haiyan/http:// http://www.theguardian.com/world/2014/mar/30/humanitarian-drones-medical-supplies-no-roads-technologyhttp:// http://www.theguardian.com/world/2014/mar/30/humanitarian-drones-medical-supplies-no-roads-technologyhttp:// http://www.theguardian.com/world/2014/mar/30/humanitarian-drones-medical-supplies-no-roads-technologyhttp://dx.doi.org/10.1016/j.pt.2014.09.001mailto:[email protected]

Opinion Trends in Parasitology xxx xxxx, Vol. xxx, No. x


influencing the transmission of infectious diseases. Land-use change, for example through deforestation or agricul-tural expansion, has been widely documented as a majordriver of infectious disease emergence and spread [14–18]. Anthropogenic environmental changes can modify thetransmission of zoonotic and vector-borne diseases by dis-rupting existing ecosystems and altering the geographicspread of human populations, animal reservoirs, and vectorspecies [19,20]. For example, the emergence of malaria innew areas of South America and Southeast Asia has beenassociated with the clearing of tropical forests resulting inchanges in anopheline mosquito densities and contact withpeople [21]. Changes in forest cover affect the life cycle anddistribution of disease vectors by altering microclimates,availability of breeding sites, and ecological communitystructures [22]. Simultaneously, deforestation is associatedwith higher levels of human activity within forest environ-ments, leading to increased exposure to forest-breedingvectors [23]. Understanding rapidly changing patterns ofhuman settlement and vector distribution in this context isvital for predicting disease risks and effectively targetingdisease-control measures.

Satellite data versus aerial dataEpidemiologists rely on accurate spatial and environmentaldata to describe variations in vector-borne and zoonoticdisease risk, establish early warning systems, model diseasetransmission, and estimate disease burden [24]. These datacan include detailed information on land cover, climaticvariables, and distributions of human and animal popula-tions. Geospatial data can be obtained from a range ofsources, such as satellite-based remote sensing, aerial sur-veys, and ground-based Global Positioning System (GPS)surveys.

Satellite remote sensing is increasingly being used toobtain environmental data on land cover, vegetation, soiltype, surface water, and rainfall for infectious disease re-search [25]. Satellite data are characterised by varyingspatial, temporal, and spectral resolutions. Temporal reso-lution relates to the frequency with which a satellite returnsto a specific location, while spectral resolution is defined bythe wavelength interval size on the electromagnetic spec-trum and the number of intervals measured by the satellite’ssensor. Higher spectral resolution allows image classifica-tion or transformation (such as for vegetation indices) usinginformation beyond the visible range of the electromagneticspectrum. A new generation of sensors such as QuickBird,IKONOS, and GeoEye (http://www.digitalglobe.com) pro-vide imagery with very high spatial resolution (

TRENDS in Parasitology

(A) (B)

Figure 1. Use of the Sensefly eBee to map land cover in Malaysia. (A) Setting up the Sensefly eBee before a flight. (B) Launching the Sensefly eBee.



and macaque movement and vector bionomics to under-stand the epidemiology of infection.

The commercially available Sensefly eBee UAV wasused for all mapping exercises (Sensefly, Cheseaux-Lau-sanne, Switzerland; Figure 1). The eBee can fly for up to50 min and uses a 16-megapixel digital camera to recordaerial images, which can be used to produce maps anddigital surface models. All UAV flight plans were pro-grammed and monitored using eMotion2 software (Sense-fly, Cheseaux-Lausanne, Switzerland) and post-flightimage processing was completed using Postflight Terra3D (Pix4D SA, Lausanne, Switzerland). ArcGIS (ESRI,Redlands, USA) was used for data analysis and generationof 3D models (Figure 2). Previews of aerial photographsand digital surface models were generated in real time,while full data processing took several hours.

Within the study sites, the eBee was flown at approxi-mately 350–400 m above the take-off point. Publicly avail-able digital elevation data from the Advanced SpaceborneThermal Emission and Reflection Radiometer GlobalDigital Elevation Model (ASTER GDEM; http://www.

Figure 2. 3D model of the stud

jspacesystems.or.jp/ersdac/GDEM/E/index.html) wasused to develop flight plans. Of 158 flights, 127 (80%)generated usable data. The most common reasons forfailed flights were high winds, rain, and battery failure.Of these flights, six (5%) were obscured by low clouds andneeded to be repeated. The average area covered by asingle flight was 124 hectares (1.24 km2) with an imageoverlap of 80–90% and an average resolution of 11.22 cmper pixel. Mapping exercises were conducted on 26 daysbetween December 2013 and May 2014, with repeatedflights over areas identified as having high rates ofland-use change (Figure 3). The resulting maps wereoverlaid with GPS data on locations of households andmalaria cases and used to characterise land-use types andcreate a spatial sampling frame for further sampling(Figure 4) [45].

Benefits of UAV mappingFor this project, spatially and temporally detailed data onthe dynamics of land use and land cover are requiredto explore interactions between environmental factors,


y site in Sabah, Malaysia.

3

http://www.jspacesystems.or.jp/ersdac/GDEM/E/index.htmlhttp://www.jspacesystems.or.jp/ersdac/GDEM/E/index.html

0 12.5 25 50 75 0

(A) (B)

12.5 25 50 75Meters Meters


Figure 3. Mapping changes to land cover at the study site in Sabah, Malaysia. (A) Study site in February 2014. (B) Same study site in May 2014 after the start of clearing to

create a rubber plantation.



disease vectors, and human and primate hosts in the lightof increasing disease transmission. As is commonly thecase in tropical settings, it proved impossible to obtainrecent cloud-free satellite data for our field sites via thearchives of satellite-data providers. Images on GoogleEarth (http://www.earth.google.com), which uses datafrom the same archives, were out of date and inadequatefor characterising the study site (for example, large-scaleclearings at one study site appeared as intact forest inGoogle Earth). The paucity of available satellite data andlack of certainty in the ability to obtain data for key time

0 15 30 60 90 120Meters

Figure 4. Macaque movements around huma

4

points moving forward were the principal reasons forelecting to conduct land-cover/land-use mapping usingan UAV.

Data from UAVs share many of the characteristics ofhigh-resolution satellite data, although in the case of UAVsthe user has more control over the spatial resolution of theresulting images (depending on flight altitude, images fromUAVs typically have a spatial resolution of 4–20 cm; cur-rently, the highest spatial resolution available from com-mercial high-resolution satellite sensors is 41 cm). As withsatellite data, UAVs can produce ‘stereo’ images that, using

Knowlesi casesKey:

Houses

Macaque movements


n Plasmodium knowlesi case household.

http://www.earth.google.com/



standard photogrammetric tools, can be used for DEMgeneration, 3D visualization, and feature extraction.

One of the main benefits of using UAVs is the ability toobtain data in real time and to repeatedly map areas ofinterest as frequently as required. In one of our sites inSabah, development began on clearing secondary forest toestablish a rubber plantation. As the clearing occurredwithin a limited geographical area, the progress of theclearing and the resulting land changes could be mappedquickly and updated routinely. This ability to map changesas they occur is critical for understanding how land-usechange affects the distribution of human populations anddisease vectors.

Depending on the size of the area to be covered and theresolution of data needed, the costs of purchasing andoperating an UAV can compare favourably with purchas-ing high-resolution satellite data over repeated timepoints. A wide range of UAV models is available, withlow-cost options like the Conservation Drone available forseveral hundred dollars and high-end, specialised dronescosting hundreds of thousands of dollars [6]. For our pur-poses, we chose a commercially available fixed-wing UAVcosting approximately US$25 000 for the UAV and associ-ated software. While less expensive models of UAV areavailable, this UAV could be easily used without signifi-cant training or technical knowledge, allowing multiplemembers of the project team to be trained in operating theUAV. Various models of UAV are commercially availablewith different specifications. The choice of an UAV modelshould depend on the financial and technical resourcesavailable, the anticipated spatial and temporal scales ofthe mapping project, and the types of data required.

Limitations of UAVsAlthough UAVs represent a new source of data for epide-miological investigations, there remain significant poten-tial limitations in their use. Similarly to light aircraft,small UAVs cannot fly in all weather conditions. Theability to withstand certain weather conditions is deter-mined by the size and specifications of the UAV used. Themodel we chose could not be used during rain or with windspeeds over 45 km/h (12 m/s). We also found that hightemperatures at our study sites (frequently in excess of358C) could cause the UAV to overheat after multiplecontinuous flights. While not as much of an issue as withsatellite data, low cloud cover can also limit the visibility ofdata collected at certain times of day or in areas with poorvisibility. The variability of weather conditions can make itdifficult to plan exact flight times ahead of time and evenwhen conditions appear suitable, areas frequently need tobe remapped to obtain sufficient data. Some land types,such as forest, are more difficult to map due to the difficultyof matching overlapping images and may need to bemapped repeatedly or at higher resolutions.

Additionally, mapping exercises using an UAV requireadequate resourcing. While small areas can be mappedquickly, mapping larger areas can require significantamounts of field personnel time. The amount of timeneeded to map an area is highly dependent on local weath-er conditions and the image resolution required. If higherresolutions of data are needed, UAVs need to be flown at

lower heights and can cover shorter distances per flight.The number of flights conducted per day may also beconstrained by the availability of electricity and abilityto recharge the UAV’s batteries. High levels of usage cannecessitate the purchase of additional equipment andspare parts as well as lead to higher maintenance costs.Further, processing and analysis of UAV data can becomputationally intensive, requiring computers with highspecifications that may not always be available in the field.While data can be processed at a later date, immediateprocessing of collected images allows rapid assessment ofdata quality and better planning of further fieldwork.UAVs have also been limited by the lack of multispectraldata, although recently UAVs have been modified to recordother data of interest; for example, UAVs have been fittedwith near-IR (NIR) cameras to measure the biomass offorest areas [46,47]. Currently, the spectral resolution ofmost UAVs is limited compared with available satellitedata; however, this is a rapidly developing technology andmay change in the near future.

Challenges can also be encountered while applying forofficial permission for conducting UAV mapping. As theuse of UAVs is relatively uncommon, there is often no clearregulatory framework for applying for permission. For ourresearch, we were required to apply for permissions frommultiple agencies, ranging from ministries of defence andcivil aviation authorities to conservation and developmentcouncils and land-use-planning authorities. While guide-lines are in place for traditional aerial surveys, theseguidelines were not always appropriate for relativelyshort, low-altitude UAV flights. For example, some regula-tions required the submission of detailed flight plans toallow redirection of other aircraft within the area, despitethe differences in flight heights between a small UAV andlarger planes. It is also worth noting that insurance asso-ciated with the use of UAVs is potentially restrictive.Although we encountered no safety issues with using anUAV, all project staff needed to be instructed on the safehandling of equipment.

Concluding remarks and future perspectivesDetailed investigations of environmental factors influenc-ing the transmission of infectious diseases are vital toeffectively target surveillance and control programmes.UAVs present a new opportunity to obtain high-resolution,georeferenced data in real time. These data can be used tobetter understand how land-use changes affect the emer-gence and spread of infectious diseases by monitoring thedistribution of human populations and changes to thehabitats of disease vectors and wildlife reservoirs. Wedemonstrated the utility of this method by using anUAV to obtain environmental data for an epidemiologicalinvestigation of risk factors for zoonotic malaria.

The use of UAVs is most appropriate when detailedmaps of relatively small geographical areas are needed inareas where high-resolution satellite data are not readilyavailable. UAVs may be inappropriate for large-scale datacollection due to the time and resources required to operatethem. Also, despite the modification of some UAVs torecord data at different wavelengths, UAVs do not havethe spectral resolution of most satellite data. Within

5



smaller areas, UAVs can be used to generate high-resolu-tion data on land cover, vegetation, and elevation and canbe used to monitor changes in habitats of vectors andwildlife reservoirs on a fine spatial scale. Additionally,UAVs can provide a valuable alternative to other datasources when data are needed either in real time or atvery frequent time points.

AcknowledgementsThe authors thank Gaim James Lunkapis (Universiti Sabah Malaysia),Tommy Rowel Abidin (Infectious Diseases Society Sabah), Albert M. Lim(Infectious Diseases Society Sabah), and Judy Dorothy Marcos (ResearchInstitute for Tropical Medicine) for their help with UAV mappingexercises. Additional thanks go to Beth Downe (London School of Hygieneand Tropical Medicine) and the teams in Sabah and Palawan for theirsupport with this project. The authors acknowledge the Medical ResearchCouncil, Natural Environment Research Council, Economic and SocialResearch Council, and Biotechnology and Biosciences Research Councilfor the funding received for this project through the Environmental andSocial Ecology of Human Infectious Diseases Initiative (ESEI). Grantnumber: G1100796.

References1 Schiffman, R. (2014) Wildlife conservation. Drones flying high as new

tool for field biologists. Science 344, 4592 Cummings, M. (2013) A drone in every driveway. Sci. Am. 308, 28–293 Mulero-Pazmany, M. et al. (2014) Remotely piloted aircraft systems as

a rhinoceros anti-poaching tool in Africa. PLoS ONE 9, e838734 Koh, L. and Wich, S. (2012) Dawn of drone ecology: low-cost

autonomous aerial vehicles for conservation. Trop. Conserv. Sci. 5,121–132

5 Getzin, S. et al. (2012) Assessing biodiversity in forests using very highresolution images and unmanned aerial vehicles. Methods Ecol. Evol. 3,397–404

6 Hodgson, A. et al. (2013) Unmanned aerial vehicles (UAVs) forsurveying marine fauna: a dugong case study. PLoS ONE 8, e79556

7 Martin, J. et al. (2012) Estimating distribution of hidden objects withdrones: from tennis balls to manatees. PLoS ONE 7, e38882

8 Pena, J.M. et al. (2013) Weed mapping in early-season maize fieldsusing object-based analysis of unmanned aerial vehicle (UAV) images.PLoS ONE 8, e77151

9 Torres-Sanchez, J. et al. (2013) Configuration and specifications of anunmanned aerial vehicle (UAV) for early site specific weedmanagement. PLoS ONE 8, e58210

10 Doherty, P. and Rudol, P. (2007) A UAV search and rescue scenariowith human body detection and geolocation. In AI 2007: Advances inArtificial Intelligence (Orgun, M. and Thornton, J., eds), pp. 1–13,Springer

11 Mendelow, B. et al. (2007) Development of e-Juba, a preliminary proofof concept unmanned aerial vehicle designed to facilitate thetransportation of microbiological test samples from remote ruralclinics to National Health Laboratory Service laboratories. S. Afr.Med. J. 97, 1215–1218

12 Pollanen, R. et al. (2009) Radiation surveillance using an unmannedaerial vehicle. Appl. Radiat. Isot. 67, 340–344

13 Corrigan, C. et al. (2008) Capturing vertical profiles of aerosols andblack carbon over the Indian Ocean using autonomous unmannedaerial vehicles. Atmos. Chem. Phys. 8, 737–747

14 Foley, J. et al. (2005) Global consequences of land use. Science 309, 570–574

15 Kilpatrick, A. and Randolph, S. (2012) Drivers, dynamics and control ofemerging vector-borne zoonotic diseases. Lancet 380, 1946–1955

16 Wolfe, N. et al. (2007) Origins of major human infectious diseases.Nature 447, 279–283

17 Patz, J. et al. (2008) Disease emergence from global climate and landuse change. Med. Clin. North Am. 92, 1473–1491

18 Jones, K. et al. (2008) Global trends in emerging infectious diseases.Nature 451, 990–993

19 Patz, J. et al. (2004) Unhealthy landscapes: policy recommendations onland use change and infectious disease emergence. Environ. HealthPerspect. 112, 1092–1098

6

20 Lambin, E. et al. (2010) Pathogenic landscapes: interactions betweenland, people, disease vectors and their animal hosts. Int. J. HealthGeogr. 9, 54

21 Guerra, C. et al. (2006) A global assessment of closed forests,deforestation and malaria risk. Ann. Trop. Med. Parasitol. 100, 189–204

22 Yasuoka, J. and Levins, R. (2007) Impact of deforestation andagricultural development on anopheline ecology and malariaepidemiology. Am. J. Trop. Med. Hyg. 76, 450–460

23 Barbieri, A. et al. (2005) Population and land use effects on malariaprevalence in the Southern Brazilian Amazon. Hum. Ecol. 33, 847–874

24 Hay, S. et al. (2012) Global mapping of infectious disease. Philos. Trans.R. Soc. Lond. B: Biol. Sci. 4, 1814

25 Hay, S. (2000) An overview of remote-sensing and geodesy forepidemiology and public health application. Adv. Parasitol. 47, 1–35

26 Machault, V. et al. (2011) The use of remotely sensed environmentaldata in the study of malaria. Geospat. Health 5, 151–168

27 Dong, J. et al. (2014) A 50-m forest cover map in Southeast Asia fromALOS/PALSAR and its application on forest fragmentationassessment. PLoS ONE 9, e85801

28 Chuang, T. and Wimberly, M. (2012) Remote sensing of climaticanomalies and West Nile virus incidence in the Northern GreatPlains of the United States. PLoS ONE 7, e46882

29 Lourenco, P. et al. (2011) Anopheles atroparvus density modeling usingMODIS NDVI in a former malarious area in Portugal. J. Vector Ecol.36, 279–291

30 Weiss, D. et al. (2014) Air temperature suitability for Plasmodiumfalciparum malaria transmission in Africa 2000-2012: a highresolution spatiotemporal predication. Malar. J. 13, 171

31 Beck, L. et al. (1997) Assessment of a remote sensing-based model forpredicting malaria transmission risk in villages of Chiapas, Mexico.Am. J. Trop. Med. Hyg. 56, 99–106

32 Dambach, P. et al. (2012) Utilization of combined remote sensingtechniques to develop environmental variables influencing malariavector densities in West Africa. Int. J. Health Geogr. 11, 8

33 Glass, G. et al. (2002) Satellite imagery characterises local animalreservoir populations of Sin Nombre virus in the southwestern UnitedStates. Proc. Natl. Acad. Sci. U.S.A. 99, 16817–16822

34 Emch, M. et al. (2010) Local population and regional environmentaldrivers of cholera in Bangladesh. Environ. Health 9, 2

35 Rogers, D. et al. (2014) Using global maps to predict the risk of denguein Europe. Acta Trop. 129, 1–14

36 Randolph, S. et al. (2000) Seasonal synchrony: the key to tick-borneencephalitis foci identified by satellite data. Parasitology 121, 15–23

37 Caughley, G. (1977) Sampling in aerial survey. J. Wildl. Manag. 41,605–615

38 Zharikov, Y. et al. (2005) Mapping and characterising subtropicalestuarine landscapes using aerial photography and GIS for potentialapplication in wildlife conservation and management. Biol. Conserv.125, 87–100

39 Reiter, M. and LaPointe, D. (2009) Larval habitat for avian malariavector Culex quinquefasciatus (Diptera: Culicidae) in altered mid-elevation mesic-dry forests in Hawai’i. J. Vector Ecol. 34, 208–216

40 Suarez, J. et al. (2005) Use of airborne LiDAR and aerial photographyin the estimation of individual tree heights in forestry. Comput. Geosci.31, 253–262

41 Asner, G. et al. (2011) A universal airborne LiDAR approach fortropical forest carbon mapping. Oecologia 168, 1147–1160

42 Erhart, A. et al. (2005) Epidemiology of forest malaria in centralVietnam: a large scale cross-sectional study. Malar. J. 4, 58

43 Lee, K. et al. (2011) Plasmodium knowlesi: reservoir hosts and trackingthe emergence in humans and macaques. PLoS Pathog. 4, e1002015

44 Singh, B. and Daneshvar, C. (2013) Human infections and detection ofknowlesi. Clin. Microbiol. Rev. 26, 165–184

45 Grigg, M. et al. (2014) Which factors are associated with the risk ofacquiring Plasmodium knowlesi malaria in Sabah, Malaysia: a casecontrol study protocol. BMJ Open 4, e006004

46 Makynen, J. et al. (2012) Multi- and hyperspectral UAV imagingsystem for forest and agriculture applications. SPIE Proc. http://dx.doi.org/10.1117/12.918571

47 Duan, S.B. et al. (2013) Land surface reflectance retrieval fromhyperspectral data collected by an unmanned aerial vehicle over theBaotou test site. PLoS ONE 8, e66972

http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0005http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0005http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0010http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0015http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0015http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0020http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0020http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0020http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0025http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0025http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0025http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0030http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0030http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0035http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0035http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0040http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0040http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0040http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0045http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0045http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0045http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0050http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0050http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0050http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0050http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0055http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0055http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0055http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0055http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0055http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0060http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0060http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0065http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0065http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0065http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0070http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0070http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0075http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0075http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0080http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0080http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0085http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0085http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0090http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0090http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0095http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0095http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0095http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0100http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0100http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0100http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0105http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0105http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0105http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0110http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0110http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0110http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0115http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0115http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0120http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0120http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0125http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0125http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0130http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0130http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0135http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0135http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0135http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0140http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0140http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0140http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0145http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0145http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0145http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0150http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0150http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0150http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0155http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0155http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0155http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0160http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0160http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0160http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0165http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0165http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0165http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0170http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0170http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0175http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0175http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0180http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0180http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0185http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0185http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0190http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0190http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0190http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0190http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0195http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0195http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0195http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0200http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0200http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0200http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0205http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0205http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0210http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0210http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0215http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0215http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0220http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0220http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0225http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0225http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0225http://dx.doi.org/10.1117/12.918571http://dx.doi.org/10.1117/12.918571http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0235http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0235http://refhub.elsevier.com/S1471-4922(14)00146-9/sbref0235

Mapping infectious disease landscapes: unmanned aerial vehicles and epidemiologyApplications of UAVsSatellite data versus aerial dataCase study: mapping environmental risk factors for zoonotic malariaBenefits of UAV mappingLimitations of UAVs

Concluding remarks and future perspectivesAcknowledgementsReferences

Documents

Mapping infectious disease landscapes: unmanned aerial … · 2015-01-16 · Mapping infectious disease landscapes: unmanned aerial vehicles and epidemiology Kimberly 1 M. Fornace1,